Image comparison is one of the most basic image processing techniques. For example, template matching which calculates the degree of similarity (or likelihood) between a target image and a template image is widely used in image recognition. Also, image compression of still images or video images often includes processing based on the degree of similarity (or degree of difference) between two images.
The degree of similarity between two images (or degree of difference between image data of two images) is most typically evaluated with a SAD (sum of absolute differences) or a SSD (sum of squared differences) of grayscale values or brightness values. For example, when two images P and Q are compared on the basis of the brightness value (Y data in the YUV format, for example) of each pixel i of images P and Q, the degree of similarity S may be calculated by the following expression (1):
                              S          =                                    ∑              i                        ⁢                          |                                                Y                  Pi                                -                                  Y                  Qi                                            |                                      ,                            (        1        )            where YPi and YQi are the brightness values of pixels i of images P and Q, respectively, and Σ represents the sum with respect to all the pixels of images P and Q. When the image data of each pixel i is given in the RGB format, the brightness values YPi and YQi of pixels i of images P and Q in expression (1) may be calculated in accordance with the following expressions (2a) and (2b), for example:YPi=0.300RPi+0.590GPi+0.110BPi,  (2a)YQi=0.300RQi+0.590GQi+0.110BQi,  (2b)where RPi, GPi and BPi are the grayscale values of the red (R), green (G) and blue (B) colors, respectively, which are indicated in the image data of image P and RQi, GQi and BQi are the grayscale values of the red, green and blue colors, respectively, which are indicated in the image data of image Q. It should be noted that various expressions may be used for conversion from RGB data to Y data as is well known in the art.
However, the degree of similarity of two images calculated on the basis of the SAD or SSD of the grayscale values of respective colors of the respective pixels or the brightness values of the respective pixels is not a parameter optimized for representing the difference between two images actually displayed on a display device (or the difference between two images actually perceived by an observer observing the display device).
In connection with this, Japanese Patent Application No. H07-231391 A discloses a technology in which a compression process selected in response to an attribute data is performed in compressing image data described with a page description language.
Japanese Patent Application Publication No. H09-200750 A discloses a block compression processing based on curved surface fitting, in which parameters of the curved surface fitting are determined on the basis of the sum of squares of compression errors.
Japanese Patent Application No. H05-155072 A discloses a technique which involves compressing a bitmap data received from a host device with a plurality of different compression schemes and selecting the one that most reduces the data amount.